河套灌区作物种植结构遥感精准识别研究  

Research on remote sensing accurate recognition of crop patterns in Hetao irrigation area

作　　者：冯雪力[1] 于欣 FENG Xueli;YU Xin(Inner Mongolia Technical College of Construction,Hohhot 010070,China;Inner Mongolia Autonomous Region Surveying and Mapping Geographic Information Center,Hohhot 010070,China)

机构地区：[1]内蒙古建筑职业技术学院,内蒙古呼和浩特010070 [2]内蒙古自治区测绘地理信息中心,内蒙古呼和浩特010070

出　　处：《经纬天地》2025年第1期59-64,共6页Survey World

摘　　要：进行作物种植结构监测分析对实现农业可持续发展具有重要意义。以河套灌区为研究区,基于覆盖作物全生育期的Sentinel-2影像时间序列数据计算多种植被指数和分析时序统计特征,并组合形成7种分类方案,分别与随机森林分类器结合,构建作物种植结构精准识别模型,以期对灌区主要农作物进行精细分类并分析最优特征。结果表明:综合全部特征的分类模型精度最高,为89.25%,略高于基于多特征的分类模型;仅考虑单一特征的分类模型精度最低,为77.7%。研究成果可为河套灌区作物分类研究监测提供数据参考。Monitoring and analysis for planting structure is of great significance to realize the sustainable development of agriculture.Taking Hetao irrigation area as an example,the Sentinel-2 image time series data covering the whole growing period of crops used in this paper calculates several vegetation indices and analyzes time series statistical characteristics to form seven classification schemes respectively,an accurate recognition model of crop planting structure was established by combining with random forest classifier,and the main crops in irrigation area were classified and the optimal characteristics were analyzed.The results show that the classification model that synthesizes all features has the highest accuracy,reaching 89.25%.It is slightly higher than the classification model based on multiple features.The classification model considering only a single feature has the lowest accuracy,which is 77.7%.The results of this study can provide data reference for crop classification research and monitoring in Hetao irrigation district.

关 键 词：时序数据 随机森林分类 决策树 农作物分类 

分 类 号：P236[天文地球—摄影测量与遥感]